And-Parallel Execution - Granul. Control
WS Presentations Main Page
Next: And-Parallel Execution -
Up: Document: tut
Previous: Concurrent Languages
- In practice there are overheads associated with parallel execution:
- Task creation and scheduling.
- Possible migration of tasks to remote processors.
- Communication overheads.
- If the granularity of parallel tasks, i.e. the ``work available''
under them, is too small, it may happen that costs are larger than the
benefits in their parallel execution.
- Aim of granularity control: change parallel execution to
sequential execution or vice-versa based on conditions related to
grain size and overheads.
- Benefits greater for systems with greater
parallel execution overheads:
- Small shared memory multiprocessors: beneficial.
- Distributed memory multiprocessors,
workstation ``farms'', etc: essential.
- Useful user support techniques identified:
- Visualization / trace analysis: e.g., VisAndOr and IDRA.
- Program transformation to increase granularity.
- Ambitious objective: construction of an automatic
granularity control system: granularity analysis + granularity
control (program transformation).
Thu Jan 18 11:21:33 MET 1996 <herme@fi.upm.es>-<webmaster@clip.dia.fi.upm.es>